Study Of The Related Technologies In Distributed Video Coding

Compared with the traditional video coding schemes, distributed video coding is a new paradigm with low encoding complexity, high decoding complexity, strong error resilience and high flexibility. And these advantages cater to the demand of the current novel multimedia, such as low-power wireless video monitoring and wireless PC camera. Due to small storage capacity and limited calculation ability, these novel multimedia need a video transmission scheme with low complexity encoder. This dissertation deeply studies the theory foundation of distributed video coding and the existing research results, and puts forward some improvement on the basis of existing research. The main research work is as follows:(1) In DVC, generating side information is an important technology. Different from the traditional video coding, side information in DVC is obtained at the decoder and is an estimation of the original WZ frame. Besides, its quality decides the rate distortion performance of DVC system. Chapter 3 maily studies the generation approaches of SI, and puts forward a side information generation algorithm based on multi-block mode decision algorithm proposed in this dissertation. First, it places a multi-block classification at the encoder which catagories the block into high motion and low one. The SI of the low motion is directly generated by MCI. But for the high motion one, its SI is optimized by the procedure proposed in the thesis. Simulation results show that compared with the traditional MCI technology, side information generation method in this dissertation is effective to improve the subjective and objective quality of the decoded image.(2) There are almost two approaches of rate control in DVC, namely rate control with and without feedback channel. In order to reduce the time-delay of decoding, rate control without feedback channel has become a research hotspot. Therefore, the thesis proposes a rate control scheme without feedback channel. At the encoder, side information estimation is based on multi-block mode decision algorithm proposed by this thesis, the correlation noise model parameters is chosen according to the movement of macro block. And finally, this thesis emploies the model parameter to calculate check rates of each bit plane. Simulation results reveal that the rate control scheme can control rate transmission well and get a higher PSNR compared with the current decoder rate control scheme.(3) Besides, the thesis puts forward an effective feedback-free distributed video coding framework based on the previous research. Side information is generated with the method proposedin chapter 3 and rate control strategy abides by the principle in chapter 4. Then, several kinds of video test sequences with different motion are selected to test the RD performance of the coding framework. The test results show that the rate distortion performance can be improved in some degree compared with the existing WZ coding scheme.